Legal claims defining the scope of protection. Each claim is shown in both the original legal language and a plain English translation.
1. A computer-implemented method for determining matching subjects for features provided in a stimulus comprising: (a) pre-processing subject information comprising: (1) receiving input, wherein the input comprises: (i) a feature type definition comprising one or more feature types, and a criterion and an assembly process to be used for each of the one or more feature types; (ii) a subject set comprising one or more subjects, and one or more features for each subject, wherein each of the one or more features comprises a combination of one of the feature types and one or more feature values; (2) for each combination of the feature type and the feature value, building a feature type-feature value BitSet across all of the one or more subjects, wherein each of the one or more subjects is represented by a bit in each feature type-feature value BitSet; (b) receiving the stimulus, wherein the stimulus comprises one or more stimulus features, wherein each stimulus feature comprises a stimulus feature type-stimulus feature value combination; (c) for each feature type in the one or more stimulus features, assembling all feature type-feature value BitSets from the subject set that match, resulting in assembled feature type BitSets; (d) building an evaluation BitSet by combining the assembled feature type BitSets; and (e) determining matching subjects that match the stimulus based on the evaluation BitSet.
This invention relates to a computer-implemented method for identifying subjects that match specific features defined in a stimulus. The method addresses the challenge of efficiently searching and matching subjects based on complex feature criteria, which is useful in applications like recommendation systems, data analysis, or user profiling. The method begins by pre-processing subject information. It receives input defining feature types, criteria, and assembly processes for each type. It also receives a subject set containing multiple subjects, each with one or more features. Each feature is a combination of a feature type and one or more feature values. The system then builds a BitSet for each combination of feature type and feature value across all subjects, where each subject is represented by a bit in the BitSet. Next, the method receives a stimulus containing one or more stimulus features, each defined by a feature type and feature value. For each feature type in the stimulus, the system assembles the relevant feature type-feature value BitSets from the subject set, resulting in assembled feature type BitSets. These BitSets are then combined to form an evaluation BitSet. Finally, the method determines which subjects match the stimulus by analyzing the evaluation BitSet, identifying subjects that meet the criteria defined in the stimulus. This approach efficiently narrows down subjects using bitwise operations, optimizing performance for large datasets.
2. The computer-implemented method of claim 1 , wherein the building the feature type-feature value BitSet comprises: building a feature type-feature value inclusion BitSet, wherein for each feature type in the combination that has the criterion of an inclusion or a range, start with a first BitSet value for every subject, and set the bit at a position that corresponds to the subject that provides the feature value in the combination to a second BitSet value that is an inverse of the first BitSet value.
This invention relates to a computer-implemented method for processing feature data, specifically for building a feature type-feature value BitSet to efficiently filter or analyze subjects based on their feature values. The method addresses the challenge of efficiently representing and querying complex feature combinations, such as those involving inclusion or range criteria, in large datasets. The method involves constructing a feature type-feature value inclusion BitSet, where each feature type in a given combination is evaluated. For feature types with inclusion or range criteria, the method initializes a first BitSet value for every subject, typically representing a default state (e.g., all bits set to 0). The method then identifies subjects that meet the specified feature value criteria and sets their corresponding bits to a second BitSet value, which is the inverse of the first (e.g., changing 0 to 1). This approach allows for efficient bitwise operations to filter or analyze subjects based on their feature values, improving performance in applications like data mining, machine learning, or database querying. The method is particularly useful for scenarios where multiple feature types must be evaluated simultaneously, such as in rule-based filtering or pattern matching, where traditional methods may be computationally expensive. By leveraging BitSet operations, the invention enables faster processing and reduced memory usage compared to alternative approaches.
3. The computer-implemented method of claim 2 , further comprising for each feature type, building a complementary BitSet, wherein each complementary BitSet has a bit that corresponds to one of the subjects.
The invention relates to a computer-implemented method for processing data, specifically for efficiently managing and querying feature data associated with multiple subjects. The method addresses the challenge of optimizing storage and retrieval of feature data, particularly when dealing with large datasets where subjects may or may not possess certain features. The core problem is reducing computational overhead and memory usage when determining which subjects have or lack specific features. The method involves organizing feature data into BitSets, where each BitSet represents a feature type and each bit within the BitSet corresponds to a subject. A bit is set (e.g., to 1) if the subject has the feature, and unset (e.g., 0) if the subject lacks it. To enhance efficiency, the method further builds complementary BitSets for each feature type. A complementary BitSet inverts the original BitSet, such that a bit is set if the subject lacks the feature and unset if the subject has it. This allows for faster queries when checking for the absence of a feature, as the complementary BitSet can be directly accessed instead of scanning the original BitSet for unset bits. The method improves performance in systems where feature presence or absence is frequently queried, such as in recommendation engines, data filtering, or machine learning pipelines. By precomputing complementary BitSets, the system avoids runtime computations for negative feature checks, reducing latency and resource consumption. The approach is particularly useful in scenarios with high-dimensional feature spaces or large subject populations.
4. The computer-implemented method of claim 3 , wherein building the complementary BitSet comprises: performing a logical OR operation of all of the feature type-feature value inclusion BitSets that have the same feature type; and obtaining the complementary BitSet by performing a logical NOT operation on a result of the logical OR operation.
This invention relates to data processing, specifically methods for efficiently managing and querying datasets using BitSets. The problem addressed is the computational overhead and complexity involved in handling large datasets with multiple feature types and values, particularly when performing inclusion or exclusion operations across these features. The method involves constructing a complementary BitSet to identify elements in a dataset that do not meet certain criteria. First, multiple feature type-feature value inclusion BitSets are generated, each representing whether elements in the dataset include a specific feature value for a given feature type. These BitSets are then combined using a logical OR operation for each feature type, aggregating all feature values under that type. The result is inverted using a logical NOT operation to produce the complementary BitSet, which identifies elements that do not include any of the specified feature values for the given feature type. This approach optimizes data filtering by reducing the number of operations required to exclude elements based on feature type-feature value combinations, improving efficiency in large-scale data processing tasks. The method is particularly useful in applications like database querying, data mining, and machine learning, where rapid filtering of datasets is essential.
5. The computer-implemented method of claim 4 , wherein the building the feature type-feature value BitSet comprises: building a feature type-feature value exclusion BitSet, wherein for each feature type in the combination that has the criterion of an exclusion, start with the second BitSet value for every subject, and set the bit at a position that corresponds to the subject that provides the feature value in the combination to the first BitSet value.
This invention relates to a computer-implemented method for processing feature data, specifically for building a feature type-feature value BitSet in a data analysis or machine learning system. The method addresses the challenge of efficiently representing and processing feature data, particularly when certain feature types have exclusion criteria that must be accounted for during analysis. The method involves constructing a feature type-feature value exclusion BitSet, which is used to filter or modify data based on exclusion rules. For each feature type in a given combination that has an exclusion criterion, the method starts with the second BitSet value for every subject (e.g., a data record or entity). The method then sets the bit at a position corresponding to the subject that provides the feature value in the combination to the first BitSet value. This operation effectively marks subjects that should be excluded or modified based on the exclusion criteria, allowing for efficient filtering or transformation of the data. The exclusion BitSet is built by iterating through the feature types and applying the exclusion rules, ensuring that the resulting BitSet accurately reflects the exclusion criteria for the given feature combination. This approach enables efficient data processing, particularly in large-scale datasets where exclusion rules must be applied dynamically. The method is useful in applications such as data preprocessing, feature selection, and machine learning model training, where exclusion criteria must be enforced to improve data quality or model performance.
6. The computer-implemented method of claim 5 , wherein building the feature type-feature value BitSet comprises: performing a logical XOR operation between the feature type-feature value inclusion BitSet and the complementary BitSet; and performing a logical AND operation between a result of the logical XOR operation and the feature type-feature value exclusion BitSet.
This invention relates to a computer-implemented method for processing feature data, specifically for constructing a BitSet that represents the inclusion and exclusion of feature types and their corresponding values. The method addresses the challenge of efficiently managing and querying large datasets where features and their values need to be dynamically included or excluded based on specific criteria. The method involves generating a BitSet that encodes both the inclusion and exclusion of feature types and their associated values. First, a logical XOR operation is performed between a feature type-feature value inclusion BitSet and a complementary BitSet. The complementary BitSet represents the inverse of the inclusion BitSet, effectively identifying features that are not included. The result of this XOR operation is then combined with a feature type-feature value exclusion BitSet using a logical AND operation. The exclusion BitSet explicitly marks features that should be excluded regardless of their inclusion status. The final output is a BitSet that accurately reflects the combined inclusion and exclusion rules for the feature types and their values, enabling efficient filtering and querying of the dataset. This approach optimizes feature selection by dynamically adjusting inclusion and exclusion criteria, improving performance in applications such as machine learning, data analysis, and database management.
7. The computer-implemented method of claim 6 , wherein the assembling all feature type-feature value BitSets from the subject set that match comprises: (a) when the criterion for the feature type is not a range, and the assembly process is mutually exclusive, using the feature type-feature value BitSet as the assembled feature type BitSet; (b) when the criterion for the feature type is not a range, and the assembly process is alternative, building and using a BitSet union that combines the feature type-feature value BitSets for feature values having a common feature type as the assembled feature type BitSet; and (c) when the criterion is a range: (1) building the BitSet union that combines the feature type-feature value BitSets for feature values within the range and having a common feature type; (i) when the range is greater than or equal to (>=), using a BitSet union of all BitSets with feature type-feature values lower than the stimulus feature type-stimulus feature value combination; (ii) when the range is less than or equal to (<=), using a BitSet union of all BitSets with feature type-feature values at or higher than the stimulus feature type-stimulus feature value combination; (2) performing a logical OR operation between a result from the above steps and the complementary BitSet, resulting in the assembled feature type BitSet.
This invention relates to a computer-implemented method for assembling feature type-feature value BitSets from a subject set based on specified criteria. The method addresses the challenge of efficiently processing and combining BitSets representing feature values of a common feature type, particularly when dealing with range-based criteria or mutually exclusive/alternative assembly processes. The method involves three primary assembly approaches. First, when the criterion for a feature type is not a range and the assembly process is mutually exclusive, the feature type-feature value BitSet is directly used as the assembled feature type BitSet. Second, when the criterion is not a range but the assembly process is alternative, a BitSet union is built by combining all feature type-feature value BitSets for feature values sharing the same feature type, resulting in the assembled feature type BitSet. Third, when the criterion is a range, the method first builds a BitSet union of feature type-feature value BitSets for values within the specified range and of the same feature type. For ranges defined as greater than or equal to (>=), the union includes all BitSets with feature type-feature values lower than a reference stimulus combination. For ranges defined as less than or equal to (<=), the union includes all BitSets with feature type-feature values at or higher than the reference stimulus combination. Finally, a logical OR operation is performed between the resulting BitSet and its complementary BitSet to produce the assembled feature type BitSet. This approach ensures accurate and efficient assembly of BitSets under various criteria and assembly conditions.
8. The computer-implemented method of claim 7 , further comprising: for each feature type that is not in the one or more stimulus features, utilizing the complementary BitSet as the assembled feature type BitSet.
The invention relates to a computer-implemented method for processing feature types in a data analysis system, particularly for identifying and handling missing or unselected feature types in a dataset. The method addresses the challenge of efficiently managing feature types that are not explicitly included in a set of stimulus features, which is critical for accurate data analysis and machine learning tasks. The method involves generating a complementary BitSet for each feature type that is not present in the one or more stimulus features. A BitSet is a compact data structure that represents a collection of bits, where each bit corresponds to a specific feature type. The complementary BitSet is derived by inverting the bits of the original BitSet, effectively marking the absence of the feature type. This complementary BitSet is then used as the assembled feature type BitSet for the missing feature types, ensuring that they are properly accounted for in subsequent processing steps. By utilizing the complementary BitSet, the method ensures that all feature types, including those not explicitly selected, are consistently represented in the data analysis pipeline. This approach improves the reliability and accuracy of feature-based analysis, particularly in scenarios where certain features may be dynamically excluded or omitted. The method is particularly useful in applications such as machine learning, data mining, and statistical analysis, where comprehensive feature representation is essential for optimal performance.
9. The computer-implemented method of claim 8 , further comprising: building the evaluation BitSet by performing a logical AND operation to the assembled feature type BitSets.
This invention relates to a computer-implemented method for evaluating features in a data processing system, particularly for optimizing feature selection or filtering in machine learning or data analysis applications. The method addresses the challenge of efficiently identifying relevant features from a large dataset by leveraging bitwise operations to reduce computational overhead. The method involves assembling multiple feature type BitSets, where each BitSet represents a binary vector indicating the presence or absence of specific feature types in a dataset. These BitSets are generated based on predefined criteria, such as feature relevance, statistical significance, or domain-specific rules. The method then builds an evaluation BitSet by performing a logical AND operation across the assembled feature type BitSets. This operation combines the BitSets to produce a final evaluation BitSet that identifies features meeting all specified criteria. The logical AND operation ensures that only features present in all input BitSets are retained, effectively filtering out irrelevant or redundant features. This approach improves computational efficiency by reducing the number of features processed in subsequent steps, such as model training or data analysis. The method is particularly useful in high-dimensional datasets where feature selection is critical for performance optimization.
10. The computer-implemented method of claim 9 , wherein from the evaluation BitSet, all positions where a bit is set to the second BitSet value correspond to the subjects that match the stimulus.
This invention relates to a computer-implemented method for identifying subjects that match a given stimulus within a dataset. The method leverages BitSet data structures to efficiently evaluate and compare data elements. The process involves generating an evaluation BitSet where each bit position corresponds to a subject in the dataset. The method then sets bits in the evaluation BitSet to a second BitSet value (e.g., 1) for subjects that meet specific criteria related to the stimulus. The resulting BitSet highlights all subjects that match the stimulus, enabling rapid identification and further processing. This approach optimizes memory usage and computational efficiency by using compact BitSet representations instead of storing full subject identifiers or complex data structures. The method is particularly useful in large-scale data analysis, pattern recognition, or filtering applications where quick and resource-efficient matching is required. By focusing on bit-level operations, the technique minimizes overhead and accelerates the matching process, making it suitable for real-time or high-throughput systems. The invention addresses the challenge of efficiently identifying relevant subjects in large datasets without sacrificing performance or scalability.
11. The computer-implemented method of claim 9 , further comprising: sorting the assembled feature type BitSets by bit count in ascending order; performing the logical AND operation to successive groups of bits within the assembled feature type BitSets in the ascending order, wherein when a resultant value of the groups of bits in the evaluation BitSet resulting from the logical AND operation is the first BitSet value: skipping performing the logical AND operation on remaining groups of bits, in the sorted assembled feature type BitSets, at a same position as the group of bits.
This invention relates to optimizing the evaluation of feature types in a computer-implemented system, particularly for improving efficiency in processing large datasets. The problem addressed is the computational overhead associated with evaluating multiple feature types, where each feature type is represented as a BitSet—a compact data structure storing binary values. The method optimizes the evaluation process by reducing unnecessary logical operations, thereby enhancing performance. The method involves assembling feature type BitSets, each representing a set of binary features. These BitSets are sorted by their bit count in ascending order, meaning those with fewer set bits are processed first. The method then performs logical AND operations on successive groups of bits within the sorted BitSets. If the result of an AND operation on a group of bits yields the first possible BitSet value (e.g., all zeros), the method skips further AND operations on remaining groups of bits at the same position in the sorted BitSets. This early termination avoids redundant computations, improving efficiency. The technique is particularly useful in applications requiring rapid feature evaluation, such as machine learning, data filtering, or pattern recognition, where minimizing computational steps is critical. By prioritizing BitSets with fewer set bits and terminating early when a trivial result is detected, the method significantly reduces processing time and resource usage.
12. A system for determining matching subjects for features provided in a stimulus comprising: (a) a server computer having a processor; (b) an application executing on the server computer that: (1) pre-processes subject information by: (A) receiving input, wherein the input comprises: (i) a feature type definition comprising one or more feature types, and a criterion and an assembly process to be used for each of the one or more feature types; (ii) a subject set comprising one or more subjects, and one or more features for each subject, wherein each of the one or more features comprises a combination of one of the feature types and one or more feature values; (B) for each combination of the feature type and the feature value, building a feature type-feature value BitSet across all of the one or more subjects, wherein each of the one or more subjects is represented by a bit in each feature type-feature value BitSet; (2) receives the stimulus, wherein the stimulus comprises one or more stimulus features, wherein each stimulus feature comprises a stimulus feature type-stimulus feature value combination; (3) for each feature type in the one or more stimulus features, assembles all feature type-feature value BitSets from the subject set that match, resulting in assembled feature type BitSets; (4) builds an evaluation BitSet by combining the assembled feature type BitSets; and (5) determines matching subjects that match the stimulus based on the evaluation BitSet.
This system is designed for identifying subjects that match specific features defined in a stimulus, addressing the challenge of efficiently searching and filtering large datasets based on complex feature combinations. The system operates by pre-processing subject information to create a structured representation of features, enabling rapid matching against a given stimulus. The server computer processes input data, which includes feature type definitions and a subject set. Each feature type definition specifies criteria and assembly processes for one or more feature types, while the subject set contains subjects with associated features, each consisting of a feature type and one or more feature values. The system builds a BitSet for each combination of feature type and feature value, where each subject is represented by a bit in the BitSet. When a stimulus is received, the system processes it by extracting stimulus features, each defined by a feature type and feature value. For each feature type in the stimulus, the system assembles matching feature type-feature value BitSets from the subject set, resulting in assembled feature type BitSets. These BitSets are then combined to form an evaluation BitSet, which identifies subjects that match all features in the stimulus. The system efficiently determines matching subjects by leveraging BitSet operations, enabling fast and scalable feature-based matching in large datasets.
13. The system of claim 12 , wherein the application builds the feature type-feature value BitSet by: building a feature type-feature value inclusion BitSet, wherein for each feature type in the combination that has the criterion of an inclusion or a range, start with a first BitSet value for every subject, and set the bit at a position that corresponds to the subject that provides the feature value in the combination to a second BitSet value that is an inverse of the first BitSet value.
This invention relates to a system for processing feature data, particularly for evaluating combinations of feature types and values in a dataset. The system addresses the challenge of efficiently determining which subjects (e.g., data entries or records) meet specific criteria based on feature types and their associated values. The system builds a BitSet structure to represent whether subjects satisfy certain inclusion or range criteria for feature types in a combination. The system constructs a feature type-feature value BitSet by first creating a feature type-feature value inclusion BitSet. For each feature type in the combination that has an inclusion or range criterion, the system initializes a BitSet with a first value (e.g., 0) for all subjects. It then sets the bit at the position corresponding to each subject that provides the feature value in the combination to a second value (e.g., 1), which is the inverse of the first value. This BitSet structure allows for efficient filtering and evaluation of subjects based on the specified criteria, enabling rapid determination of which subjects meet the defined conditions. The approach optimizes performance by leveraging BitSet operations, which are computationally efficient for large datasets.
14. The system of claim 13 , wherein for each feature type, the application builds a complementary BitSet, wherein each complementary BitSet has a bit that corresponds to one of the subjects.
This invention relates to a system for efficiently managing and querying feature data associated with multiple subjects, such as documents, users, or other entities. The problem addressed is the computational inefficiency of traditional methods when handling large-scale feature data, particularly when performing operations like filtering or searching across numerous subjects. The system organizes feature data into feature types, where each feature type represents a distinct category of attributes. For each feature type, the system constructs a complementary BitSet, a data structure where each bit corresponds to a subject. The complementary BitSet is designed such that a bit is set (e.g., to 1) if the corresponding subject does not possess the feature, and unset (e.g., to 0) if the subject does possess the feature. This inversion allows for efficient logical operations, such as AND, OR, or NOT, to be performed directly on the BitSets, significantly reducing the computational overhead compared to traditional methods that process individual feature values. The system further includes a mechanism to dynamically update the BitSets as new subjects or features are added or removed, ensuring the data structure remains accurate and up-to-date. The use of complementary BitSets enables rapid filtering of subjects based on feature presence or absence, making the system particularly useful in applications requiring real-time or near-real-time querying, such as recommendation systems, document retrieval, or user profiling. The approach optimizes memory usage and processing speed, making it scalable for large datasets.
15. The system of claim 14 , wherein the application builds the complementary BitSet by: performing a logical OR operation of all of the feature type-feature value inclusion BitSets that have the same feature type; and obtaining the complementary BitSet by performing a logical NOT operation on a result of the logical OR operation.
This invention relates to data processing systems that handle feature type-feature value pairs, particularly for efficiently building complementary BitSets in database or information retrieval applications. The problem addressed is the computational overhead and complexity involved in generating BitSets that represent the absence of certain feature-value combinations, which is useful for filtering or querying large datasets. The system includes a method for constructing a complementary BitSet, which is a data structure indicating which elements do not match a given feature type and value. The process involves first generating multiple feature type-feature value inclusion BitSets, each representing the presence of a specific feature-value pair. These BitSets are then logically combined using an OR operation for all BitSets sharing the same feature type. The result of this OR operation is inverted using a NOT operation to produce the complementary BitSet, which identifies elements that do not have the specified feature type and value. This approach optimizes the generation of exclusion-based BitSets by leveraging logical operations on precomputed inclusion BitSets, reducing the need for repeated calculations and improving efficiency in systems that rely on feature-based filtering or querying. The method is particularly useful in applications where rapid exclusion checks are required, such as in database indexing, machine learning feature selection, or information retrieval systems.
16. The system of claim 15 , wherein the application builds the feature type-feature value BitSet by: building a feature type-feature value exclusion BitSet, wherein for each feature type in the combination that has the criterion of an exclusion, start with the second BitSet value for every subject, and set the bit at a position that corresponds to the subject that provides the feature value in the combination to the first BitSet value.
This invention relates to a system for processing feature data, particularly for efficiently managing and querying combinations of feature types and values. The system addresses the challenge of handling large datasets where features (attributes) and their values need to be dynamically filtered or excluded based on specific criteria. The system builds a BitSet structure to represent feature type-feature value combinations, optimizing storage and retrieval operations. The system constructs a feature type-feature value BitSet by first generating a feature type-feature value exclusion BitSet. For each feature type in a combination that meets an exclusion criterion, the system starts with the second BitSet value for every subject (e.g., a data record or entity). It then sets the bit at a position corresponding to the subject that provides the feature value in the combination to the first BitSet value. This approach ensures that excluded feature values are properly marked, allowing efficient filtering during queries. The exclusion BitSet helps in dynamically adjusting the feature combinations based on predefined rules, improving performance in applications like data analysis, machine learning, or database management. The system enhances data processing by reducing computational overhead and enabling faster exclusion-based filtering.
17. The system of claim 16 , wherein the application builds the feature type-feature value BitSet by: performing a logical XOR operation between the feature type-feature value inclusion BitSet and the complementary BitSet; and performing a logical AND operation between a result of the logical XOR operation and the feature type-feature value exclusion BitSet.
This invention relates to a system for managing feature types and their associated values in a data processing environment. The system addresses the challenge of efficiently tracking which feature types and values are included or excluded in a dataset, particularly in applications where dynamic filtering or feature selection is required. The system includes a mechanism for generating a BitSet that represents the inclusion and exclusion of feature types and their values. The BitSet is constructed by performing a logical XOR operation between an inclusion BitSet (indicating which feature types and values are included) and a complementary BitSet (representing the inverse of the inclusion BitSet). The result of this XOR operation is then combined with an exclusion BitSet (indicating which feature types and values are explicitly excluded) using a logical AND operation. This process ensures that only the feature types and values that are both included and not excluded are retained in the final BitSet. The system is designed to optimize performance by minimizing computational overhead during feature selection, particularly in large-scale data processing tasks. The use of BitSet operations allows for efficient bitwise manipulation, reducing the need for iterative checks or complex conditional logic. This approach is particularly useful in applications such as machine learning, data filtering, or real-time analytics where rapid feature selection is critical. The system ensures that the feature type-feature value BitSet accurately reflects the desired inclusion and exclusion criteria, enabling precise and efficient data processing.
18. The system of claim 17 , wherein the application assembles all feature type-feature value BitSets from the subject set that match by: (a) when the criterion for the feature type is not a range, and the assembly process is mutually exclusive, using the feature type-feature value BitSet as the assembled feature type BitSet; (b) when the criterion for the feature type is not a range, and the assembly process is alternative, building and using a bitset union that combines the feature type-feature value BitSets for feature values having a common feature type as the assembled feature type BitSet; and (c) when the criterion is a range: (1) building the BitSet union that combines the feature type-feature value BitSets for feature values within the range and having a common feature type; when the range is greater than or equal to (>=), using a BitSet union of all BitSets with feature type-feature values lower than the stimulus feature type-stimulus feature value combination; (ii) when the range is less than or equal to (<=), using a BitSet union of all BitSets with feature type-feature values at or higher than the stimulus feature type-stimulus feature value combination; and (2) performing a logical OR operation between a result from the above steps and the complementary BitSet, resulting in the assembled feature type BitSet.
The invention relates to a system for assembling feature type-feature value BitSets in a data processing environment. The system addresses the challenge of efficiently combining BitSets representing different feature values of a common feature type, particularly when dealing with criteria that may involve ranges or non-range conditions. The system processes a subject set of data, where each feature type-feature value BitSet corresponds to a specific feature value within a feature type. The assembly process can be either mutually exclusive or alternative, depending on the application's requirements. For non-range criteria, the system either directly uses the feature type-feature value BitSet (mutually exclusive) or combines relevant BitSets into a union (alternative). For range-based criteria, the system constructs a BitSet union for values within the specified range, adjusting for >= or <= conditions by including all lower or higher values, respectively. The result is further refined by performing a logical OR operation with the complementary BitSet to produce the final assembled feature type BitSet. This approach ensures accurate and efficient assembly of BitSets under various conditions, improving data processing performance.
19. The system of claim 18 , wherein for each feature type that is not in the one or more stimulus features, the application utilizes the complementary BitSet as the assembled feature type BitSet.
The invention relates to a system for processing and analyzing feature data, particularly in the context of machine learning or data analysis pipelines. The system addresses the challenge of efficiently handling feature types that are not present in a given set of stimulus features, which is critical for reducing computational overhead and improving processing speed. The system includes a mechanism for assembling feature type BitSets, which are data structures used to represent the presence or absence of specific features in a dataset. For feature types that are not included in the stimulus features, the system uses a complementary BitSet as the assembled feature type BitSet. This approach leverages the complementary BitSet, which is a precomputed or dynamically generated BitSet that represents the inverse of the stimulus features, ensuring that all non-stimulus features are accounted for without redundant computations. The system further includes a feature extraction module that processes input data to identify and extract relevant features, and a feature assembly module that organizes these features into BitSets for efficient storage and retrieval. The use of complementary BitSets for non-stimulus features optimizes memory usage and speeds up subsequent operations, such as feature selection or model training, by avoiding unnecessary processing of irrelevant features. This invention is particularly useful in applications where feature sets are large or dynamically changing, such as real-time data analysis, natural language processing, or image recognition, where efficient feature handling is essential for performance.
20. The system of claim 19 , wherein the application builds the evaluation BitSet by performing a logical AND operation to the assembled feature type BitSets.
A system for evaluating data features in a computing environment addresses the challenge of efficiently identifying relevant features from large datasets. The system includes a processor and memory storing instructions that, when executed, perform operations to process feature data. The system constructs a BitSet for each feature type, where each BitSet represents the presence or absence of a specific feature in the data. These BitSets are assembled based on input data, allowing for rapid comparison and evaluation. The system then builds an evaluation BitSet by performing a logical AND operation on the assembled feature type BitSets. This operation combines the individual BitSets to determine which features are present across all evaluated data, enabling efficient feature selection and analysis. The system may further include a user interface for displaying the evaluation results, allowing users to visualize and interact with the processed data. The logical AND operation ensures that only features present in all input data are included in the final evaluation, improving accuracy and reducing computational overhead. This approach is particularly useful in applications requiring high-speed feature extraction, such as machine learning, data mining, or real-time analytics.
21. The system of claim 20 , wherein from the evaluation BitSet, all positions where a bit is set to the second BitSet value correspond to the subjects that match the stimulus.
This invention relates to a system for evaluating subject responses to a stimulus, particularly in the context of data processing and pattern matching. The system addresses the challenge of efficiently identifying subjects that match a given stimulus by leveraging bitwise operations for rapid comparison. The system includes a first BitSet representing a stimulus and a second BitSet representing a set of subjects. Each bit in the BitSet corresponds to a specific attribute or characteristic, where a set bit indicates the presence of that attribute. The system evaluates the subjects by performing a bitwise comparison between the stimulus BitSet and each subject BitSet. This comparison generates an evaluation BitSet, where each bit position corresponds to a subject. A bit in the evaluation BitSet is set to a second BitSet value (e.g., 1) if the subject matches the stimulus, and to a different value (e.g., 0) otherwise. The system then processes the evaluation BitSet to identify all subjects that match the stimulus by extracting the positions where the bit is set to the second BitSet value. This approach enables efficient and scalable matching, particularly useful in applications requiring rapid filtering or classification of subjects based on their attributes. The system may be applied in fields such as data analysis, machine learning, or user profiling, where identifying matches between stimuli and subjects is critical.
22. The system of claim 20 , wherein the application further: sorts the assembled feature type BitSets by bit count in ascending order; performs the logical AND operation to successive groups of bits within the assembled feature type BitSets in the ascending order, wherein when a resultant value of the groups of bits in the evaluation BitSet resulting from the logical AND operation is the first BitSet value: skips performing the logical AND operation on remaining groups of bits, in the sorted assembled feature type BitSets, at a same position as the group of bits.
This invention relates to a system for efficiently processing feature type BitSets in a data analysis or machine learning context. The problem addressed is the computational inefficiency in evaluating multiple BitSets, particularly when performing logical AND operations across large datasets. The system optimizes this process by sorting the assembled feature type BitSets by bit count in ascending order, then performing logical AND operations on successive groups of bits within these sorted BitSets. When the result of a logical AND operation yields the first BitSet value, the system skips further operations on remaining groups of bits at the same position, improving processing speed and resource utilization. The system also includes a method for assembling feature type BitSets from input data, where each BitSet represents a feature type and its corresponding values. The logical AND operations are performed in ascending order of bit count, ensuring that the most restrictive (lowest bit count) BitSets are evaluated first, which allows early termination of the evaluation process when a match is found. This approach reduces unnecessary computations and enhances efficiency in feature evaluation tasks.
Unknown
October 1, 2019
Browse 5M+ US patents with plain-English claim translations and AI-generated analysis.